The European Central Bank published a working paper developing an endogenous-threshold vector autoregression framework for cases where the variable determining economic regimes moves contemporaneously with system shocks. The method uses a regime-specific copula control function to remove this dependence without external instruments or assumptions about the threshold variable’s marginal distribution, while retaining linear least-squares estimation within each regime. The paper finds that conventional models can overstate both the initial impact and subsequent propagation of shocks when threshold endogeneity is ignored. Monte Carlo exercises show that the correction substantially reduces bias in estimated thresholds, regime-specific covariance matrices and impulse responses across Gaussian, heavy-tailed and nonlinear dependence structures. Applied to U.S. monetary transmission from 1973 to 2007, the model estimates an inflation threshold of about 3.5%, with the high-inflation regime covering 48.55% of the sample. A normalized 25-basis-point tightening produces faster disinflation and a lower estimated sacrifice ratio in the high-inflation regime, about 1.6 compared with 5.0 in the low-inflation regime, while the linear VAR generates a price puzzle and obscures these differences.
European Central Bank2026-07-29
European Central Bank working paper develops endogenous-threshold VAR correction for monetary transmission analysis
A European Central Bank working paper develops a copula-based threshold VAR that corrects for contemporaneous dependence between regime indicators and shocks. Simulations show lower bias in thresholds, covariance estimates and impulse responses. In a U.S. application, monetary tightening produces faster disinflation and a lower sacrifice ratio when inflation is above an estimated threshold of about 3.5%.